An Introduction to Regression Discontinuity Design and Potential for Application in Neurology.

Abstract

Regression discontinuity design (RDD) offers a rigorous approach for estimating causal effects using observational data where randomized clinical trials are not feasible by leveraging cutoff-based treatment rules, effectively accounting for confounding when certain assumptions hold true. RDD remains underused in neurologic research, with most existing applications in neurology emerging only in the past 5 years. We introduce RDD and explain how effects can be estimated within this framework, including a discussion of the key assumptions required for valid causal inference. In addition, we highlight the relevance and potential applications of RDD in neurologic research, particularly in contexts where treatment decisions are based on clinical or policy thresholds. We also outline best practices and limitations associated with this method, with the aim of encouraging thoughtful applications of RDD in neurologic research.